Synthetic and Augmented Transposable Element Datasets for Classification and Curation Benchmarks

Directory structure and dataset contents

The Classification directory contains three FASTA files:

r.1.5_all_dataaug.fasta:Database including real and synthetic sequences, where synthetic data were generated using data augmentation techniques.

r.1.5_all_GAN_GROUPED_calibrated_augmented.fasta:Database including real and synthetic sequences, where synthetic data were generated using a grouped GAN approach (two GANs: one trained on Class I TEs and one on Class II TEs).

r.1.5_all_GAN_SIMPLE_calibrated_augmented.fasta:Database including real and synthetic sequences, where synthetic data were generated using a single, simple GAN model.

The Curation directory contains two FASTA files:

synthetic_dataset_with_imperfections.fasta:Database including real and synthetic sequences, generated across five experimental scenarios, where synthetic sequences were further modified by introducing imperfections and fragmentations at levels of 0.15 and 0.3.

synthetic_dataset_without_imperfections.fasta:Database including real and synthetic sequences, generated across the same five experimental scenarios, without introducing imperfections or fragmentations.